HP530 Overheating issue

I've followed this guide and successfully replaced the thermal paste on my HP8510p. It went completely smooth. And the core temperatures dropped for an enourmous 50°C (the laptop was completely clogged with dust and to make it even worse the factory-applied thermal paste was a mess). Anyway, it's been idling at around 55°C and it never ever goes past 65°C (3D gaming and all).

I've just done this to a friend's notebook. The weird thing is, that it was filled with dust and idling at around 80-85°C. I took it apart following this guide. After I removed all of the plastic and the dust and the fan and the heatsink, I used the ArctiClean set to clean the cores and the heatsink, applied a nice thin layer of ArcticSilver and put it all back together again, to find that it's still idling at 75-80°C. I know that the CPU is a Intel Centrino Merom-2M (@1,60GHz) and that supposedly mobile CPUs run at higher temps... But this?
I took it apart again, redid everyhing, this time even more meticulously and still nothing. Whenever I do a CPU hungry task (compressing some random files with winrar for instance) the temps go all the way up to 84°C.
Now here's the weird thing:
When my 8510p is hot, the air coming from the fan is also quite hot. When the 530 is at 84°C the air coming from the fan is room temp. And it won't even return to below 70°C if left idling.
(I know, I did everyhing right (the heatsink screws surrounding the CPU are numbered and I did each one little by little, according to the rest), but it seems that the heat just isn't transferring from the CPU to the heatsink.)

I am completely stumped.
Any suggestions anyone?

[For details about the notebook check the second guide]

EDIT: The temps do drop under 70°C, but only if left idling. If a video is playing, 70°C.

 

If the processor is hot but the air blown out by the cooler is cold, the heatsink isn't fitted properly. Maybe there is too much space between heatsink and CPU, try inserting a copper shim between them and see if things get better.

 

check the fan speed ..it may not work properly
and use other temp reading software for comparisson, in case of false reading

 

Well, the thing is, that the fan speed change accordingly to the temps. So I figure that the readings and speeds are correct. Except if the CPU isn't heating up at all and the sensor failed completely.
The heatsink is fitted properly. I'm sure, because when I was taking it apart the second time (undoing my first try) the square from the CPU was nicely visible in the layer of paste left on the heatsink.
I'll try some other software... I'm not sure what to do. I could blame it on the sensor and lower the fan speeds just to keep it on the bearable side... But then that could lead to more problems...
And since I've taken it all apart I'll have a hard time with warranty. (I didn't want to return it in the first place, because these people are idiots. At least in my parts. They will blame you for the dust and charge you 30€ for cleaning the computer and when you take it back again they will say that you're doing everything wrong and that the comp is not to be blamed... I've seen it happen.)

 

If the heatsink fits properly, in the square the CPU leaves in the thermal paste there should be almost no paste left, the bare metal of the heatsink should be exposed.

Just try inserting a shim and let us know how it goes.

 

Well yes. The bare metal is exposed. There is a thin layer of paste left, obviously since it stuck to the CPU and the heatsink.
I don't think the shim would be a good idea - too much pressure on the core.

I've noticed some very weird behaviour.
Some screenshots:

The difference in temp reading between the two programs is HUGE! Btw, these two progs act normal on my laptop, reporting the same temps.

And another one:

The temps during load. Again, HUGE difference.
I'm thinking that the sensors are to blame.

Ideas?

 

Try CoreTemp or RealTemp.

As for "too much pressure", i've had boards that would bend almost 1cm due to the force of the cooler. Nothing bad ever happened to them or the CPUs.

 

Will do.
I'll report back tomorrow, since I had to return the lappy just now - it's not a problem, they live right next door.

 

So sorry for the late reply...

Well I tried the progs you mentioned and this is what I got:





As you can see, the CPU temp sensor failed a couple of times.
And the temps still don't match by far.

Any more suggestions? Please?

 

The reason why the temps don't match is that the two programs have different junction temperatures that they base their readings upon. You see, Intel chips don't report temperatures as an absolute value, but rather as a negative delta which represents the distance to maximum allowed temperature. And not every software maker reads all Intel docs...

While i've always found CoreTemp to be accurate (while everybody else swore by Real Temp), there is no way a mobile chip has a Tjunction of just 85C so it is wrong. And if it is indeed running in the high 90s then yes you have a problem. However, if it is worth fixing, that's another question.

I'll give you a little story which may answer your concern. I used to have a Pentium Dual-Core E2180 in my main computer a couple years ago. That thing ran quite well even heavily overclocked, but one thing was bugging me. The idle temperatures were mid 20s (basically room temp), but load registered in the high 70s - with the heatsink COLD, and yes i was using an aftermarket cooler, the Scythe Mugen which i still have today. I tried everything, from lapping (sanding the CPU's heatspreader down to a mirror finish) to a screw-based cooler retention system instead of those crappy push pins, and finally removing the heatspreader altogether.

The end result? Temps stayed EXACTLY THE SAME. I was so annoyed with that chip that i stopped the fan on the cooler and left it running Orthos. The result? CoreTemp reported 125C along with a (?) symbol, and THE HEATSINK WAS STILL COLD. Funny thing is, the chip still ran just fine. I have since replaced it with an E6550, but it's still sitting in a drawer somewhere and i believe it's still functional.

The bottom line? Unless fan noise is an issue, i'd leave it as it is. Mobile chips don't have a heatspreader anyway, so that can't be an issue here.

 

Unfortunately, fan noise IS the issue.
It got me poking around the laptop in the first place. Then, I discovered high temps that stayed the same even after replacing the thermal paste and removing a truck load of dust from the cooling system...
Anyway, I got the idea of lowering the fan speeds. I did it on my 8510p and it's near silent when idling. Worked great. But I'd have to find a way to extract the ACPI settings for the 530 since I have a neat .net program for my 8510p...

What I'd like to know is:
Would lowering the fan speeds (they are on full blast most of the time) be a sensible thing to do?
Keeping in mind that frying the cpu is not an option.

 

It won't fry... It may slow down or shut down though.